A blade server is a flat server and comprises only a central processing unit (CPU) and a system controller. The blade server can be effectively incorporated into a large data center. A small control box is able to store ten or more vertically disposed blade servers. The control box comprises an intelligent system for providing relevant information to an internal resource store and for separating network based blade server resource from specific tasks or application programs being run on the blade servers. As a result, the use of the blade servers in the control box has a great flexibility.
The blade servers can be interconnected by means of a universal switch and can share a common element (e.g., integrated switch). In addition to the blade servers, the control box further comprises an internal control module for managing all available system resources in any time and assigning tasks to a specific blade server. While the above purpose can be achieved by several conventional servers, the employment of blade servers provides advantages of having an integrated network architecture, elements for supporting hot plug, and being capable of managing the architecture. All of the above advantages form a flexible platform for operation, make a future expansion easy and a high utilization possible, and reduce the cost of buying a server.
Many companies and internet service providers (ISPs) use several tens or hundreds low cost blade servers for meeting the needs of accessing the Internet. A blade server only occupies the space of a control box unit in a data center and has height of 1.75″ complying with a standard set by the art. The blade server is also called an IU server. The blade server can be packed in the control box as books arranged on a bookshelf. More CPUs are arranged in a single control box so as to establish a high density operating environment. This is because the CPU of blade server can consume less power and generate less heat as compared to the CPU of a well known server. Moreover, a heat dissipating and power system can be shared. Thus, a high efficiency still can be achieved in such small, operation-based architecture. Also, each control box has a set of network cable and power cord for facilitating cable and cord management. Moreover, all blade servers and other components in the control box can be replaced if such need arises. As an end, the maintenance of the whole platform is made easy.
Additionally, the most important characteristics of the blade server are high reliability and high expansibility. This is because there are many blade servers in the control box. The blade server itself has a backup capability in operation. Also, the control box can provide a high utilization because it can support hot plug of the blade server and system elements. Another blade server can immediately replace a malfunctioned one without interrupting service. Moreover, the blade servers in the control box can be grouped so that a user can set levels of backup depending on needs. For example, insertion of more blade servers into the control box can improve the total performance of the control box and optimize resource distribution.
In view of the above, the blade server can bring a great profit in a lowest potential risk. In other words, the blade server has the following benefits: (1) It has a high expansibility and reliability because the control box can support hot plug of the blade server and system elements. (2) Low power consumption and shared heat dissipating device. (3) Compactness. (4) It can separate system resource from specific tasks or application programs being run. (5) Low cost.
Thus, it is desirable for a distributed blade server arrangement to provide following features as improvements of the blade server continue. These features comprise dynamically providing a powerful operational capability to each application program as required, and a more advanced, simple management for coping with unexpected load variation and peak load, increasing efficiency of system manager, decreasing the possibility of human errors, effectively operating equipment, lowering power consumption, determining whether a power supply can provide a sufficient power to the whole system in any of different combinations of blade servers, controlling the switching of each blade server for optimizing power supply, and prolonging a useful life of the power supply.